ANALYTlCALBKXHEMISTRY151,369-374(1985)

Mechanism of Dye Response and Interference

in the Bradford Protein Assay’

STEVE J. COMPTON~ AND CLIVE G. JONES~

Institute of Ecosystem Studies, New York Botanical Garden, Mary Flagler Cary Arboretum,
Box AB, Millbrook, New York 12545
Received March 7, 1985

Bradford Coomassie brilliant blue G-250 protein-binding dye exists in three forms: cationic,
neutral, and anionic. Although the anion is not freely present at the dye reagent pH, it is this
form that complexes with protein. Dye binding requires a macromolecular form with certain
reactive functional groups. Interactions are chiefly with arginine rather than primary amino groups;
the other basic (His, Lys) and aromatic residues (Try, Tyr, and Phe) give slight responses. The
binding behavior is attributed to Van der Waals forces and hydrophobic interactions. Assay in-
terference by bases,detergents, and other compounds are explained in terms of their effectsupon
the equilibria between the three dye forms. o 1985 Academic PW, I~C.
KEY WORDS: Coomassie brilliant blue G-250; Bradford protein assay; spectrophotometry;
binding assays;peptides; detergents.

The protein determination method of examining the spectroscopy and chemistry of

Bradford ( 1) is popular because it is rapid, the dye and assay responses to various poly-
sensitive, relatively inexpensive, and specific amino acids, amino acids, and natural prod-
for protein. This specificity has been particu- ucts. Our objectives were to establish the re-
larly emphasized empirically (l-4) but little quirements for, and factors influencing bind-
effort has been made to investigate the mech- ing of CBBG.
anism of Coomassie brilliant blue G-250
(CBBG)4 dye binding, or the identities of the MATERIALS AND METHODS
functional groups involved. Fazekas de St. Chemicals (Sigma, Aldrich) were of the best
Groth et al. (5) suggested that dye-binding was grade available. Solutions were made with
due to electrostatic attractions of the dye’s sul- deionized water ( 18 MQ) or dimethyl sulfoxide
fonic groups (Fig. 1) with protonated primary (DMSO). CBBG dye reagent concentrate
amino groups of protein, such as lysine and mixed according to Bradford (1) (Bio-Rad
N-terminal groups. We tested this model by Laboratories) was diluted fivefold with water
and stored at 5 “C in opaque glass. The reagent
was stable for at least 30 days, based on re-
’ Financial support was provided by the Mary Flagler sponse to bovine serum albumin (BSA) and
Cary Charitable Trust, The American Philosophical So- ribulose 1,5-diphosphate carboxylase oxygen-
ciety (74 Hays), The Research Corporation (9843), and
the National Science Foundation (DEE8 1179 113). ase (RUDP) standards. Spectra and absor-
* Present Address: University of Washington School of bances were measured with a Pet-kin-Elmer
Medicine, WAMI E-85, T-545 Health Sciences Center, Model 552 dual-beam spectrophotometer with
Seattle, Washington 98 195. recorder, using disposable 10 X 10 X 45-mm
3 To whom correspondence should be addressed. polystyrene cuvettes (Fisher). Test samples
4 Abbreviations used: CBBG, Coomassie brilliant blue
G-250; DMSO, dimethyl sulfoxide; RUDP, ribulose 1,5- listed within the text were aqueous, 10 mM (or
diphosphate carboxylase oxygenase; BSA, bovine serum 1.0 mg/ml polymer solutions). Adenine and
albumin; SDS, sodium dodecyl sulfate. indole were prepared in DMSO and compared

369 0003-2697185 $3.00

Copyright 0 1985 by Academic F’res, Inc.
All rights of reproduction in any form reserved.
370 COMPTON AND JONES

described as red and blue forms. However, a

series of dye solutions varying in pH (0.30-
1.25) and SDS concentration (O-0.3%) showed
no isobestic points in the spectral region stud-
ied, 190-750 nm. Had only two dye forms
:NH been present, two or more isobestic points
would have been seen (11). We therefore set

43 bEt
out to determine the number and identity of
CBBG dye species by measuring visible spectra
as we titrated the dye reagent in the absence
FIG. 1. Structure of Coomassie brilliant blue (CBBG) of protein. We reasoned that as protonation
G-250 dye. of the anionic dye species (Fig. 1) produced
the neutral and cationic dye species succes-
sively, modifications of the conjugated system
against DMSO solvent blanks. Polycytidylic would change the dye absorbance maximum.
acid and polyguanylic acids were 5’, Kf salts. The dye reagent absorbance spectrum
The molecular weight of polyguanylic acid was showed maxima at 470 and 650 nm (Fig. 2).
> 100,000 (Sigma), and that of polycytidylic Stepwise additions of sodium hydroxide de-
acid undetermined. Test samples were assayed creased absorbance at 470 nm with increased
using the macro method of Bradford (1) (100 absorbance at 650 nm, then replaced the 650-
~1 of test solution was mixed with 5.0 ml nm peak with a new peak at 595 nm (Fig. 2a).
CBBG dye reagent). The mean of three rep- Addition of concentrated sulfuric acid to the
licate solvent blanks was subtracted from the reagent led to replacement of the peak at 650
mean of three replicate samples. Absorbance nm with a single maximum at 470 nm (Fig.
readings were taken between 10 and 15 min 2b). The dye structure requires that the 595
to eliminate temporal variation in dye re- nm absorbing species be the unprotonated dye
sponse ( 1). A Fisher Model 6 10A Accumet pH
meter with A. H. Thomas (Philadelphia, Pa.)
combination electrode No. 4094-L25 was used
for pH determinations.

RESULTS AND DISCUSSION

According to the model of Fazekas de St.
Groth et al. (5), a reduction in the number of
protein primary amino groups should result
in reduced dye response. We found, however,
that treatment of protein in conditions favor-
ing covalent modification of protein lysine
residues (6, 7) had no significant (P = 0.05)
effect upon dye response (RUDP treated with
1.0 M p-benzoquinone, 0.1 N NaOH, 1 h,
25 “C). We therefore investigated factors influ-
encing dye binding with the aim of identifying
Wavelength (nm)
the responsible functional groups.
FIG. 2. Dye spectra. (-) Bradford dye reagent; (---) dye
pH and Dye Spectra reagent treated with: (a) NaOH, pH 1.25; (b) H$O,, pH
0.30; (c) 140 pg BSA, pH 0.78; (- . -) same, blank subtracted;
Previous workers (3, 8- 10) have supposed (d) (---) dye reagent treated with excess SDS, pH 0.77;
that CBBG dye exists as two species, usually solid line indicates 595 nm.
 RESPONSE AND INTERFERENCE IN THE BRADFORD PROTEIN ASSAY 371

anion. The successive appearance of the 650- caused by the second protonation is an effect
and 470-nm species with decreasing pH allows of the decreased extent of conjugation.
their identification as the singly and doubly
protonated species, with neutral and + 1 Dye-Protein Spectra
charge, respectively:
We next wished to establish which of the
three dye species binds with protein. Addition
OH- OH- of protein to the dye reagent results in a
ANION H+I NEUTRAL ~1’ CATION.
marked decrease of the 470-nm absorption, a
SPECIES new shoulder at 595 nm, and little change at
595 nm (blue) 650 nm (green) 470 nm (red) 650 nm (Fig. 2~). Subtraction of the dye blank
results in a single absorption maximum at 595
nm (whose magnitude is a function of the pro-
Our data are in agreement with previous stud-
tein concentration) as well as negative minima
ies (12) of a similar trianilinomethane dye,
at 470 and 650 nm. The three different ab-
Crystal Violet (N-hexamethyl-p-trianilino-
sorbance maxima of the dye reagent-protein
methyl chloride). The conjugated system of
mixture correspond to the three species of dye
Crystal Violet is almost identical to that of
present: the doubly protonated, positively
CBBG, and its pH-dependent color changes
charged species absorbing at 470 nm, the singly
(from violet to green to yellow with decreasing
protonated neutral species at 650 nm, and the
pH) are well known (13,14). pKb values for
dye-protein complex at 595 nm. Based on the
the first and second protonations of Crystal
identical absorbance maxima of the dye-pro-
Violet are 1.80 f 0.03 and 1.15 + 0.07, re-
tein complex and the dye anion, we conclude
spectively (12). We have found (Fig. 2) that
that the bound dye species is in fact the dye
CBBG is almost entirely doubly protonated at
anion. The equilibria shown above are forced
pH 0.30, and unprotonated at pH 1.25, im-
to the left as the anion is bound by protein.
plying pKb values between these extremes. The
lower pKb values for CBBG are probably due
Functional and Structural Requirements of
to the influence of its less basic secondary ni-
Protein for Dye Binding
trogen [cf. the pKb of diphenylamine is 0.79,
while the pKb of N,N-dimethylaniline is The assay’s specificity for protein is dem-
5.15 (15)]. onstrated by its lack of response to a wide range
Although the absorbing species are readily of chemical classes, including a diversity of
identified by charge, the protonation sites re- nitrogenous compounds. Polycytidylic acid,
main to be established. The dye form shown polyguanylic acid, adenine, purine, pyrimi-
(Fig. 1) is but one of several resonance struc- dine, mandelonitrile, nicotine, and caffeine
tures, and the uncharged tertiary nitrogen is were all found to give essentially no response
equivalent to that shown carrying the positive to the assay (595-nm mean blank corrected
charge. We suggest that the first protonation absorbance < 0.0 11 AU). Although the assay’s
occurs at one of the two tertiary amines, fol- specificity has been demonstrated, no studies
lowed by protonation at the other. The sul- have been made of the binding requirements
fonic acid groups are less susceptible to pro- of CBBG. To identify the protein functional
tonation than the amino groups, due to their group(s) responsible for dye binding, we tested
strong acidities [cf. the pK, of benzosulfonic the response of the Bradford assay to a variety
acid is 0.70; that of naphthalenesulfonic acid of synthetic polyamino acids. The greatest re-
is 0.57 (15)]. The bathochromic shift caused sponse (by eightfold) was given by poly-L-ar-
by the first protonation is due to the charge ginine, followed by the aromatic and other ba-
increase along the x band of the conjugated sic polyamino acids (Table 1). The low reac-
system (16), and the hypsochromic shift tivity of poly-L-lysine (17) (Table 1) is not
372 COMPTON AND JONES

TABLE 1 threshold has been suggested for the binding

DYE-BINDING RESPONSES TO of CBBG dye in both the Bradford system (3)
POLYMERIC L-AMINO ACIDS and in staining electrophoretic gels (18, 19).
Since polyglycine produced no dye response,
Mean blank the peptide chain alone cannot be responsible
corrected
Polymeric Molecular absorbance Relative for dye binding. We conclude that a com-
amino acid weight (595 nm) response 0 pound must have both a macromolecular form
and an active basic or aromatic functional
PoMArg) b 40,000 4.033 36.0 group in order to bind with the CBBG dye
PMTW 100,000 0.531 4.7 anion. Our data also suggest that the Bradford
pol~(TryY 5,000 0.491 4.4
11,ooo 0.472 4.2 assay will be most responsive to arginine-rich
poMW
PoWhe) 15,000 0.215 1.9 proteins, as suggested by electrophoretic gel
wlyVw,Ala) staining with CBBG ( 10).
loo,0 35,000 0.112 1.0
80,20d 38,000 0.028 0.3 Binding Forces
46,54d 37,000 0.001 0.0
0,100’ 25,000 0.000 0.0 The binding properties of CBBG may best
WIYWY) 6,000 0.000 0.0
be explained by the binding energies involved.
poly(Asn) 9,000 0.000 0.0
20,000 0.000 0.0
While the ionized sulfonic groups of the dye
poly(Asp)
may well experience electrostatic interactions
Note. Samples are aqueous, 1.O mg/ml unless otherwise with positively charged functional groups of
stated. protein, the response of the aromatic polypep-
’ Compared to poly(lysine). tides demonstrates the importance of non-
b Sample was diluted to I25 pg/ml to bring value in
range. The value shown is dilution corrected. electrostatic interactions. Fredericq (20, 2 1)
’ Dissolved in DMSO, 1.O mg/ml. estimated the relative energetic contributions
dRandom copolymers of L-alanine and L-lysine, by of various organic anion functional groups to
molar ratio. protein binding, based on binding studies of
BSA with anions similar to CBBG. Using these
values, we estimated the relative contributions
compatible with the model of Fazekas de St.
of ionic and nonionic forces to the CBBG-
Groth et al. (5). Furthermore, dye response to
protein complex (Table 2). The nonionic
the uncharged aromatic polyamino acids binding energies would account for about 58%
demonstrates that nonelectrostatic interac-
of the total binding energy, and help to explain
tions must play a role in Coomassie blue dye
the requirement for a lengthy peptide, as well
binding.
as observed dye binding by nonionic polypep-
We also assayed free L-amino acids (argi- tides such as poly+tyrosine.
nine, histidine, lysine, tryptophan, alanine,
glycine, asparagine, and aspartic acid) and
Interferences
functional group moieties corresponding to the
reactive polypeptides (indole for tryptophan; An important drawback of the Bradford
phenol for tyrosine; imidazole for histidine; n- method stems from its variation in response
butylamine for lysine; agmatine for arginine to different proteins, as discussed elsewhere ( 1,
and lysine; guanidine hydrochloride for argi- 22, 23). For any given protein, however, fur-
nine, and 4-hydroxybenzoic acid for tyrosine). ther discrepancies may arise from nonprotein
None of these compounds gave a significant interferences that produce protein overesti-
color response at concentrations equivalent to mation, underestimation, and a reduction of
the polyamino acid functional group concen- the linear response range. These sources of in-
trations (595 nm mean blank corrected ab- terference are best explained from the knowl-
sorbance < 0.006 AU). A molecular weight edge of the ionic dye species involved. In the
 RESPONSE AND INTERFERENCE IN THE BRADFORD PROTEIN ASSAY 373

TABLE 2 binding process. All dye-flavonoid mixtures

CALLXJLATED ENERGY CONTRIBUTIONS TO BINDING OF giving a response had absorbance maxima at
C~~MASSIE BRILLIANT BLUE G-250 DYE BY PROTEIN, 650 nm, with no peaks or shoulders observed
BASED ON RELATIVE VALUES OBTAINED BY FREDERICQ at 595 nm. Lastly, although addition of suf-
(I 1,12) WITH BOVINE SERUM ALBUMIN AND ORGANIC ficient base will generate the free dye anion
ANIONS SIMILAR TO CBBG
(Fig. 2a) through dye equilibrium shifts ac-
Structufal Free energy companying pH changes, basic protein buffers
component Frequency in increment are generally too dilute to do so. However,
(Energy, cal/mol) CBBG dye (cal/mol) even dilute basic buffers can generate neutral
dye, which interferes as described above. For
Sulfonate group this reason, we feel that the use of a buffered
(3500) 2 7000
Benzene ring dye reagent may substantially improve the
(1ow 6 6000 assay.
-CH*- (350) 10 3500 Underestimation of protein may result from
factors reducing dye binding. In addition to
stabilizing the neutral dye species, some de-
absence of interfering components, the protein tergents are known to decrease assay response
assay measures the amount of dye anion (595 to a number of enzymes ( 1,2, 19), most likely
nm) present as compared to a dye reagent- due to competition with the dye for protein.
solvent blank. Since the concentration of free The net result of these opposing effects is dif-
anion is negligible at the reagent pH of 0.77, ficult to predict. Competitive effects leading to
the response must be due to dye anion bound protein underestimation have been demon-
to protein. The blank-subtracted 595-nm ab- strated with guanidine hydrochloride (3 1) and,
sorption is proportional to the concentration in our laboratory, sodium ascorbate. Com-
of bound anion, provided that the free cation petition effects of known added reagents may
is present in sufficient excess to generate fur- generally be compensated through their inclu-
ther anion. sion in the standard calibration.
Overestimation of protein will occur when
a nonprotein produces a 595-nm absorption TABLE 3
that is not compensated for by the solvent- DYE-BINDING RESPONSES TO FLAVONOID COMPOUNDS
dye blank. Proteins appear to be the only AND SDS, MEASURED AT THE PROTEIN A~.~AY
compounds capable of generating the dye an- WAVELENGTH (595 nm)
ion at the assay pH. However, a number of
compounds are capable of favoring stabiliza- Molecular Mean blank corrected
Compound weight absorbance (595 nm)
tion of the neutral species. Since the bandwidth
of the 650-nm neutral species produces strong Flavone 222.2 0.052
absorption at 595 nm, this stabilization does Rutin 610.5 0.089
indirectly cause interference. Quercetin 302.3 0.676
This interference is seen in the case of de- Quercetrin 484.2 0.568
Kaempferol 286.3 0.743
tergents (Fig. 2d), flavonoids, and basic protein
Apigenin 270.2 0.968
buffers. The response of the Bradford assay to Chrysin 254.2 0.930
detergents permits its use in the study of de- Myricetin 318.2 0.683
tergent strength (24). Although the protein as- F&tin 286.2 0.724
say has been used in the presence of detergents SDS 288.4 0.130
(25,26), prior removal (27-29) may improve Note AI1 compounds were dissolved in DMSO (10 mh%)
sensitivity and linearity. Certain flavonoid except SDS ( 10 mM in water). Actual dye absorption max-
compounds give the same color reaction as ima were at 650 nm. Absorption maxima of llavone, rutin,
SDS (30) (Table 3), possibly through a direct and SDS were confirmed at higher concentrations.
374 COMPTON AND JONES

While the potential interference problems 3. Sedmak, J. J., and Grossberg, S. E. (1977) Anal.
of over- and underestimation due to additives Biochem. 79, 544-552.
may generally be avoided using proper con- 4. Spector, T. (1978) Anal. Biochem. 86, 142-146.
5. Fazekas de St. Groth, S., Webster, R. G., and Datyner,
trols, an important side effect is a reduction A. (1963) Biochim. Biophys. Acta 71, 377-391.
in the assay’s linear response range (2,4). As- 6. Pierpont, W. S. (1969) Biochem. J. 112, 609-617.
say response to protein is due to the equilib- 7. Pierpont, W. S. (1969) Biochem. J. 112, 619-629.
rium shift away from the undetected cation, 8. Bradford, M. M., and Williams, W. L. (1977) U. S.
through the neutral species, to the anion ab- Patent No. 4,023,933.
9. Read, S. M., and Northcote, D. H. (198 1) Anal.
sorbing at 595 nm. Since the linear assay re- Biochem. 116, 53-64.
sponse must therefore be due to the net con- 10. Reisner, A. H., Nemes, P., and Bucholtz, C. (1975)
version of dye cation to anion, the concentra- Anal. Biochem. 64, 509-5 16.
tion of free cation in the reagent ultimately II. Tinoco, I., Sauer, K., and Wang, J. C. (1978) Physical
Chemistry: Principles and Applications in Biolog-
limits linear range. Compounds stabilizing the
ical Sciences. pp. 433-435, Prentice-Hall, Engle-
neutral species do so at the expense of the dye wood, N. J.
cation pool. Although the additional neutral 12. Adams, E. Q., and Rosenstein, L. (1914) J. Amer.
absorption may be compensated for by using Chem. Sot. 36, 1452-1473.
a dye blank, the overall result is a smaller pool 13. Pratt, L. S. ( 1947) The Chemistry and Physics of Or-
of dye cation, and a reduction in the linear ganic Pigments, p. 17, Wiley, New York.
14. Venkataraman, K. (1952) The Chemistry of Synthetic
range of the assay. Dyes, Vol. 1, p. 356, Academic Press, New York.
While many workers consider the Bradford 15. Weast, R. C. (ed.) (1975) The CRC Handbook of
assay the method of choice due to its sensitiv- Chemistry and Physics, p. D-147, CRC Press,
ity, specificity, and speed, its limitations must Cleveland.
also be considered. The assay’s specificity for 16. Lewis, G. N., and Calvin, M. (1939) Chem. Rev. 25,
273-328,
arginine residues is a probable cause of vari- 17. Lea, M. A., Grasso, S. V., Hu, J., and Sidler, N. (1984)
ability in response between proteins, and re- Anal. Biochem. 141, 390-396.
agents affecting the dye equilibria can cause 18. Righetti, P. G., and Chillemi, F. (1978) J. Chromatogr.
interference. We hope that this more mecha- 157,243-25 1.
nistic understanding of the Bradford reagent 19.Saleemuddin, M., Ahmad, H., and Husain, A. (1980)
Anal. Biochem. 105.202-206.
system will allow anticipation of other possible 20. Fredericq, E. (1954) Bull. Sot. Chim. Belg. 63, 158-
interferences and promote a more effective use 181.
of the assay. 21. Fredericq, E. (1955) Bull. Sot. Chim. Be/g. 64,639-
657.
Note added in proof Ahmad and Saleemuddin (1985) 22. Pierce, J., and Suelter, C. H. (1977) Anal. Biochem.
(Anal. Biochem. 148,533-541) found that covalent mod- 81,478-480.
ification of BSA lysine groups by glutaraldehyde had little 23. Van Kley, H., and Hale, S. M. (1977) Anal. Biochem.
effect on dye response. 81,485-487.
24. Rosenthal, K. S., and Koussale, F. (1983) Anal. Chem.
ACKNOWLEDGMENTS 55,1115-1117.
25. Macart, M., and Gerbaut, L. (1982) Clin. Chim. Acta
We thank K. Marx, J. D. Hare, T. Hess, J. Mahoney, 122,93-101.
C. Parravano, J. J. Sedmak, and P. J. Silk for critical com- 26. Rubin, R. W.. and Warren, R. W. (1977) Anal.
ment. This communication is a part ofthe Chemical Ecol- Biochem. 83,773-777.
ogy program and a contribution to the program of the 27. Duhamel, R. C., Meezan, E., Brendel, K. (1981) J.
Institute of Ecosystem Studies, The New York Botanical Biochem. Biophys. Methods 4, 73-80.
Garden. 28. Zaman, Z., and Verwilghen, R. L. (1979) Anal.
Biochem. 100,64-69.
29. Kapp, 0. H., and Vinogradov, S. N. (1978) Anal.
REFERENCES Biochem. 91,230-235.
30. Godshall, M. A. (1983) J. Food Sci. 48, 1346-1347.
1. Bradford, M. M. (1976) Anal. Biochem. 12,248-254. 31. Bearden, J. C. (1978) Biochim. Biophys. Acta 533,
2. Bio-Rad Laboratories (1984) Bio-Rad Protein Assay. 525-529.